This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: GCC's -fsplit-stack disturbing Mach's vm_allocate
- From: Svante Signell <svante dot signell at gmail dot com>
- To: Samuel Thibault <samuel dot thibault at gnu dot org>
- Cc: Thomas Schwinge <thomas at codesourcery dot com>, Ian Lance Taylor <iant at google dot com>, gcc-patches at gcc dot gnu dot org, bug-hurd at gnu dot org, Roland McGrath <roland at hack dot frob dot com>
- Date: Tue, 06 May 2014 10:06:49 +0200
- Subject: Re: GCC's -fsplit-stack disturbing Mach's vm_allocate
- Authentication-results: sourceware.org; auth=none
- References: <87fvln6jjp dot fsf at schwinge dot name> <20140416220345 dot GZ5545 at type dot youpi dot perso dot aquilenet dot fr> <20140418080311 dot GA5626 at type dot bordeaux dot inria dot fr> <1398328750 dot 568 dot 74 dot camel at G3620 dot my dot own dot domain> <20140501224530 dot GD5592 at type dot youpi dot perso dot aquilenet dot fr> <1399017803 dot 8487 dot 5 dot camel at PackardBell-PC> <1399018692 dot 8487 dot 7 dot camel at PackardBell-PC> <20140502095753 dot GC6577 at type> <20140502100027 dot GD6577 at type> <1399027556 dot 8487 dot 12 dot camel at PackardBell-PC> <20140502105234 dot GM6577 at type>
- Reply-to: svante dot signell at gmail dot com
On Fri, 2014-05-02 at 12:52 +0200, Samuel Thibault wrote:
> Svante Signell, le Fri 02 May 2014 12:45:56 +0200, a écrit :
> > On Fri, 2014-05-02 at 12:00 +0200, Samuel Thibault wrote:
> > > Samuel Thibault, le Fri 02 May 2014 11:57:53 +0200, a écrit :
> > > > So we just need to fix guardsize in our libpthread.
> > >
> > > (And I'll have a look at it).
> >
> > Maybe this can fix the around 40 segfaults (of 50 failures) when split
> > stack is disabled too? The segfaults are always around the same place in
> > libpthread.
>
> Possibly, if libgo uses small stacks already.
Status update:
For some (yet) unknown reason all libgo tests fails with a segfault when
run in the build tree: make, sh or something else, the test commands are
rather hard to track.
However, running the tests manually, sigh!, after installing gccgo and
libgo5 from gcc-4.9-4.9.0-1 with split stack support the result is as
follows:
Total: 122
PASS: 95
FAIL: 27
Number of tests succeeding with libpthread's stack guard fixed: 33 (and
the other libc fixes)
Note: In many of the failing tests, only a small subset of each test
fails.
GNU/Hurd should now to be considered for upstream inclusion?
Updated patches will be sent again to gcc-patches for review.